Access Null pointer - M4 vs M33

[ari_v2]

Hello,

Code that was compiled for stm32F407 ( core M4) has an bug that accesses null pointer. The assembly code is as follows:

STRB  r1, [r0, #0x01]

Where r0=0 and r1=0.

Although MCU accesses null pointer, it does not enter hard-fault.

But, the same code on stm32U585 (core M33, w/o trust zone), runs into hard-fault.

Is there a way to make the stm32U585 to behave as the stm32F407, in this aspect.

This code cannot get changed and it must run on both MCUs.

Any Ideas?

Thank you,

Ari

[Tesla DeLorean]

There's the FPB and you can also return from Hard Faults..

[ari_v2]

Thanks for reply.

Could you give me a link how to implement a safe return from hardfault for M33 core?

[Tesla DeLorean]

The LR call-gate indicates which stack frame the context is pushed on too. Check the faulting opcode, and advance the PC

[Pavel A.]

But this means change in the code

[Tesla DeLorean]

Ok, pretty sure F4 and U5 have entirely different memory maps. I'm throwing paint in the air, hoping some of it will stick somewhere.

My guess would be there is *some* flexibility, and the issue here is some library, object or blob, where source is unavailable, then again it's not like that stuff can't get patched.

Let's say I can move the VTOR, and manipulate a vector, or two..